Activity5_Joining.Rmd

---
title: "Movies and Ratings"
author: "Nathan Diekema"
date: "9/30/2021"
output:
  prettydoc::html_pretty: 
    theme: leonids
---
## Setup 


```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE, warnings=FALSE)
```

```{r libraries, message=FALSE, warning=FALSE}
library(tidyverse)
library(reprex)
```


## The Data
**Movie_Ratings.csv**

Download link:  https://www.dropbox.com/s/ebr2gzy95pb9lsx/Movie%20Ratings.csv?dl=1 

Variables:

* Film - The title of a film.
* Genre - The film's primary genre.
* Rotten Tomatoes % - The average movie rating from critics on Rotten Tomatoes, from 0-100.
* Audience Ratings % - The average movie rating voted by audience members on Rotten Tomatoes, from 0-100.
* Budget (million $) - The operating budget of the movie.
* Year of release - The year the movie hit theaters.
 

**imdb_1000.csv**

Download link: https://www.dropbox.com/s/ov5cntaof9lj9v6/imdb_1000.csv?dl=1 

Variables:

* star_rating - The rating of the movie from user votes on imdb.com.  (0-10)
* title - The title of the film
* content_rating - The designation of the movie from the Motion Picture Association.
* genre - The primary genre of the film.
* duration - The length of the film, in minutes
* actors_list - The leading actors credited in the movie.

## The Tasks

#### Cleaning/Plotting

*1.  Read in and summarize the data.*

```{r message=FALSE, warning=FALSE}

imdb <- read_csv("data/imdb_top_1000.csv")

imdb <- imdb %>%
  separate(Genre,
           sep=',',
           into=c("Genre1","Genre2","Genre3")) %>%
  mutate(
    Runtime = parse_number(Runtime)
  )

head(imdb)

imdb %>% 
  summarize(
    Num_Movies = n_distinct(Series_Title),
    Num_Genres = n_distinct(c(Genre1,Genre2,Genre3)),
    Avg_Runtime = mean(Runtime),
    Avg_IMDB_Rating = mean(IMDB_Rating),
  )

```

*2.  What genre had the highest average imdb rating?*

```{r}

imdb_long <-
  imdb %>% 
  pivot_longer(
    c(Genre1,Genre2,Genre3),
    names_to = "Genre Number",
    values_to = "Genre"
  ) %>% 
  drop_na(IMDB_Rating)

imdb_ratings <-
  imdb_long %>% 
  group_by(Genre) %>% 
  summarize(
    mean_rating = mean(IMDB_Rating)
  ) 
  
imdb_ratings %>% 
  arrange(desc(mean_rating))



```

As you can see from the table above, **Western** movies had the highest average IMDB rating.


*3.  Is there a relationship between the content rating of the movie (e.g. "PG-13") and its duration?  Make a plot.*

```{r fig.align='center'}
imdb %>% 
  drop_na(Certificate) %>% 
  ggplot(aes(x = as.character(Certificate), y = Runtime)) +
    geom_bar(stat="summary", color="black", fill="lightsalmon1")

```

**Conclusion:**  
From the chart above it can be asserted that there is not a strong relationship between a movies content rating and it's run time. The most notable trend is the notably higher run time of unrated movies.


#### Pivoting

*1. Make a column plot comparing Rotten Tomato critic and audience ratings for all the Romance movies.*

```{r message=FALSE, warning=FALSE, fig.align='center'}
ratings <- read_csv("data/Movie_Ratings.csv")
head(ratings)

ratings_long <-
  ratings %>% 
  drop_na(c("Rotten Tomatoes Ratings %", "Audience Ratings %")) %>% 
  pivot_longer(
    c("Rotten Tomatoes Ratings %", "Audience Ratings %"),
    names_to = "rating_type",
    values_to = "rating"
  )

ratings_long %>% 
  filter(Genre=="Romance") %>%
  ggplot(aes(x=rating_type, y=rating)) +
    geom_bar(stat="summary", color="black", fill='aquamarine3')

```


*2. For each year, find the average audience rating difference between Comedy and Drama movies.*

```{r, fig.align='center', fig.width=8}

ratings_comedy_drama <- ratings_long %>% 
  filter(Genre==c("Comedy", "Drama"))

ratings_comedy_drama %>% 
  ggplot(aes(x=as.factor(`Year of release`), y=rating, fill=Genre)) +
    geom_boxplot() +
    xlab("Genre") +
    ylab("Rating")

```

The average audience rating when compared between Comedy and Drama movies differed the most in 2007 and the least in 2009.


#### Joining

*1. How many movies appear in both datasets?*

```{r message=FALSE, warning=FALSE}

count(
  semi_join(x=ratings, y=imdb, by=c("Film"="Series_Title"))
  )

```

The datasets given have **41** movies in common.


*2. How many movies appear in only the imdb dataset?*

```{r message=FALSE, warning=FALSE}

count(
  anti_join(x=imdb, y=ratings, by=c("Series_Title"="Film"))
)

```

The IMDB dataset contains **959** unique movies that are not in the Rotten Tomatoes dataset. 

*3. How many movies appear in only the Rotten Tomatoes dataset?*

```{r message=FALSE, warning=FALSE}

count(
  anti_join(x=ratings, y=imdb, by=c("Film"="Series_Title"))
)

```
The Rotten Tomatoes dataset contains **521** unique movies that are not in the imdb dataset.

## Joining and pivoting

*Make a plot comparing the ratings from Rotten Tomatoes Critics, Rotten Tomatoes Audience, and imdb.*

```{r fig.width=8, fig.align='center', message=FALSE, warning=FALSE}

# Clean data sets
imdb_clean <- 
  imdb %>% 
  drop_na(IMDB_Rating) %>% 
  mutate(
    IMDB_Rating = IMDB_Rating * 10
  )

tomato_clean <-
  ratings %>% 
  drop_na(c("Rotten Tomatoes Ratings %", "Audience Ratings %"))

all_ratings <- 
  full_join(x=imdb_clean, y=tomato_clean, by=c("Series_Title"="Film")) %>% 
  pivot_longer(
    c("Rotten Tomatoes Ratings %", "Audience Ratings %", "IMDB_Rating"),
    names_to = "rating_type",
    values_to = "rating"
  )

all_ratings %>% 
  ggplot(aes(x=as.character(rating_type), y=rating, fill=rating_type)) +
    geom_boxplot() +
    ylab("Rating") +
    xlab("Rating Type") +
    labs("Rating Type")

```

The box-plot above provides a comparison between the distribution of ratings from IMDB, and rotten tomatoes audience and critics ratings. I scaled IMDB by 10 so it was consistent with the rotten tomatoes percentage scale out of 100%. IMDB varies the least and and has the highest mean rating. The Rotten tomato ratings are a little less forgiving and vary significantly more. The Audience ratings are generally higher than the critics ratings and vary less.